Anandtech: AMD-ATI Merger in the Works?

[ERK]

Isn't this still like fourth or fifth-hand (the [H])? Might it not be that intel is just putting to rest, to partners and customers, concerns that were brought up by the rumors and Charlie's story?

Despite my doubts, I'm kinda rooting for this to happen--spice up things for a while...

 

[Geo]

I don't know how many hands down, but Kyle did not say he heard it directly from Intel, and yeah that makes a difference. Tho he must have found it reasonably credible or he wouldn't have put it on the front page.

And it's always possible it's all a giant misunderstanding from that initial report. Somebody at Intel figured they'd get asked, so had an answer for why it wouldn't hurt them anyway without any idea if there was substance to it. . .and this turned into "Intel is telling their customers" by the time it gets out a few generations of telling.

I certainly hope that's it. I don't want to see either major graphics IHV becoming junior partner to anybody.

 

[BrynS]

While non-committal, Kristopher Kubicki from DailyTech confirms in a blog entry that the Computex chattering classes were somewhat rapt by the rumour and definately did with the chattering!

He also has a not unexpected R600 morsel and some G80 info which I would place (with all the authority I can muster as an armchair 3D industry analyst with...um, 3 years of Beyond3D er...experience, yeah - now where's my certificate of competence ) in the "very likely misconstrued powerpoint" file. :smile:

 

[Arun]

He also has a not unexpected R600 morsel and some G80 info which I would place (with all the authority I can muster as an armchair 3D industry analyst with...um, 3 years of Beyond3D er...experience, yeah - now where's my certificate of competence ) in the "very likely misconstrued powerpoint" file. :smile:

Nah. I'd be very surprised if it wasn't just NVIDIA's misinformation department having some fun of their own

Uttar

 

[Brimstone]

Slightly off topic, but I think fitting with everyone liking to analyze this stuff. Over at the Gaming-Age forum, a virtual stock market game has been set-up. It's totally free, and it takes less than a minute to register for it. With the possible merger of ATI with AMD it should be an intresting 30 days.

Just click on this link to sign up. Start buying now, and then on Monday when trading starts your order will be executed.

http://vse.marketwatch.com/Game/StartViewGame.aspx?id=GAFGame

Password: gamingage
Starts: 6/11/06
Ends: 7/11/06
Starting Cash: $100,000.00

The game starts monday (stock market opens), so we won't see any results until atleast Monday. Go ahead and sign up, set up some buys and the game will kick off monday!

So far the game has 10 people signed up, hopefully some people from Beyond3d join in. Of course you can invest into any stock, it doesn't have to be tech releated.


Link to the Gaming-Age forum thread.

http://www.neogaf.com/forum/showthread.php?t=105448

 

[KimB]

Just found this piece that ars technica did on this rumor a couple of days ago:

http://arstechnica.com/articles/culture/amd-ati.ars

 

[Geo]

Interesting article, but again everyone seems to want to focus on why AMD would want to do this deal, and less so on why ATI would agree it is in their longterm interest as well.

 

[KimB]

Well, it's more important as to whether the board of directors at ATI would agree to it. They are likely to care more about money than the company's products, and a big fat payoff may well be just what the doctor ordered.

 

[karlotta]

Well, it's more important as to whether the board of directors at ATI would agree to it. They are likely to care more about money than the company's products, and a big fat payoff may well be just what the doctor ordered.

Im not sure how the shares are held/who has the biggest lot, but i think ATI can be had on a hostile. And the the companys products are what makes them (directors) money. ATi as a company doesnt need AMD at all. But the shareholders might want a new value. I beleive they have to vote on this. Not just hav D.O. give the keys to Austin.

 

[Dave Baumann]

The Street.com has has an article on AMD, and has something at the end concerning these rumours, however they last line may be one of the most salient points:

Jelinek points to another rumor as having some bearing. Were AMD to acquire graphics chipmaker ATI Technologies (ATYT:Nasdaq - commentary - research - Cramer's Take), as one Wall Street analyst predicted last month, an additional fabrication facility would not be so excessive.

"You can put a scenario together that says with additional mergers and acquisitions at AMD, then potentially, it would be a situation where they would be able to fill that [fab capacity] relatively quickly," says Jelinek.

True, ATI is a fabless chip company, meaning that it designs its graphics processors itself and outsources the manufacturing to a third party. But AMD's business model of doing most manufacturing in house means the company could conclude that it's more cost-effective to manufacture the graphics processors itself in the event of an acquisition.

That said, many people consider a marriage between AMD and ATI to be highly unlikely, given the two companies' different product focus and business models.

For such strange bedfellows, a shared fab might be the only thing they have in common.

Its been suggested to me by some analysts that some kind of tie up along the fabbing lines would make a lot more sense than a merger itself; also bear in mind that Rich Heye, former fairly big man at AMD is now at ATI, so they probably have fairly good lines of communication for such matters.

http://www.thestreet.com/_yahoo/tech/semis/10292823.html?cm_ven=YAHOO&cm_cat=FREE&cm_ite=NA

 

[Geo]

Interesting idea. So some type of joint operating agreement where ATI promises certain volumes per year and AMD promises certain capacity per year? One assumes for leading edge processes, and ATI's other work (say chipsets) stay with other fabs?

Another thing to consider. Docherty is a fab guy. His presence might make such a switch from pure fabless both more attractive internally at ATI at senior levels (i.e. a senior-level advocate), and provide more confidence for the rest of the exec team that they can knoweldgeably participate/oversee such an arrangement.

John Docherty
Senior Vice President, Operations


John Docherty is Senior Vice President of Operations for ATI Technologies Inc. In this role Mr. Docherty has overall responsibility for ASIC operations and supply chain management for ATI.

Prior to joining ATI, Mr. Docherty was Vice President of Front-End Operations for Agere Systems. In this capacity, Mr. Docherty oversaw all of Agere’s external sourcing, internal production and joint venture operations for procuring integrated circuits. Prior to joining Agere in 2002, he was Senior Vice President of Manufacturing Operations for Chartered Semiconductor Manufacturing in Singapore, where he doubled capacity and drove culture change and performance improvement across Chartered’s fab operations.

During his 19-year career with the Semiconductor Products Sector of Motorola, Mr. Docherty held various management positions in operations, culminating as Vice President of European Die Manufacturing responsible for wafer fab activity at Motorola’s, East Kilbride, (Scotland), South Queensferry, (Scotland), and Toulouse MOS 20 (France) locations. Among his accomplishments were developing accountability processes, introducing, transferring and ramping process technologies to new facilities, and implementing major cost improvement programs. While in Scotland, he participated in the establishment of the Scottish Electronics Forum and was appointed the inaugural chairman.

 

[Voltron]

Its been suggested to me by some analysts that some kind of tie up along the fabbing lines would make a lot more sense than a merger itself; also bear in mind that Rich Heye, former fairly big man at AMD is now at ATI, so they probably have fairly good lines of communication for such matters.

Even so, why would ATI be interested in a JV? Capital expenses for new fabs are huge, and that is not in either ATI's or NVIDIA's business model. If capacity is a concern, thats what TSMC and others are there for. Sure, supply can be an issue at times, but no semiconductor companies can perfectly balance supply and demand. Alternatively, would AMD make back the money it would spend on a fab by selling chips to ATI at prices competitive with the likes of TSMC? No. CPU prices and margins are much higher than graphics.

 

[Geo]

Alternatively, would AMD make back the money it would spend on a fab by selling chips to ATI at prices competitive with the likes of TSMC? No. CPU prices and margins are much higher than graphics.

Well, that wouldn't be the point for AMD. The point would be to provide more capacity for themselves without bearing the full cost --not to find a new profit stream.

 

[Jawed]

ATI seems to have a recent history of pushing processes quite hard along with the Fast-14 and X-architecture licensing agreements. And with the R520 accident last year and ATI seemingly having to second-source 80nm production before 80nm products even hit the market, a "friendly fab" could be quite compelling: presumably they'd get faster turn-around (at decent prices).

Jawed

 

[Voltron]

Well, that wouldn't be the point for AMD. The point would be to provide more capacity for themselves without bearing the full cost --not to find a new profit stream.

I'm just saying if its not a JV, where ATI ponies up A LOT of capital (which is not consisatent with their relatively low margin business model), the alternative would be for ATI to act as a de facto foundry and sell ATI chips. And I highly doubt it whether AMD could sell it to them at high enough prices to justify its own huge investment.

 

[Geo]

I dunno, what is the yearly capacity of such a fab, and what percentage would ATI need? 10% or 90%? It probably makes a difference. I dont' see any reason to think that even JV requires 50/50.

 

[BrynS]

If a JV is in the offing and not a strategic alliance alongs the lines of leveraging ATI's chipset R&D and IP together with AMD's fabrication capability/space, I would wonder why AMD isn't pursuing nVIDIA instead, assuming the equity requirement isn't overbearing and that nV didn't turn down the offer?

As Jawed mentioned, since R3xx at least, ATI have appeared to be more open to second-sourcing fab partners while pushing process nodes and die size, although nVIDIA have used a number of partners other than TSMC before (IBM, UMC, STMicroelectronics?), albeit less aggressively.

I can understand why AMD would be interested in buying ATI rather than nVIDIA outright, but insofar as a JV with ATI rather than NV -- why? The cosy AMD/NV relationship, NV's massive AMD chipset share, NV's cash reserves, etc -- why pursue ATI at all when nVIDIA already has it's proverbial foot in the door? As Dave mentioned, (an) ex-AMD executive(s) may be making the case within ATI, but that can't be the only "in" that AMD would rely on and no doubt a lot of networking and relationships have been nurtured between AMD and NV executives over the years, notwithstanding the likelyhood that high-level personnel have transitioned between both organisations too.

If AMD gets someone to spread their debt for a new fab as well as a guaranteed customer for x million dies, what carrot do they give to ATI (apart from access to very good fabs) to entertain a deal that could potentially constrain ATI's growth, while likely encouraging their rival into the (much larger) hands of Intel?

 

[Jawed]

I have to admit I'm utterly mystified why people think that there's some existing relationship twixt AMD and NVidia. All I see is that NVidia decided to produce AMD-exclusive chipsets - perhaps on the basis back then that they wouldn't be competing against Intel.

What makes people think that there's more going on?

Jawed

 

[Dave Baumann]

As Jawed mentioned, since R3xx at least, ATI have appeared to be more open to second-sourcing fab partners while pushing process nodes and die size, although nVIDIA have used a number of partners other than TSMC before (IBM, UMC, STMicroelectronics?), albeit less aggressively.

AFAIK thats been going on far longer than post R300; its more the case that with R300, the reduction in overall companies and the change press changes bought on by the internet these types of details are making it more into the public domain.

However, should there be some fab arrangment, or something like that, does it need to go as far as a joint venture? Could it be more along the lines of an agreement on effective customer fabbing in order to soak up any surpluss capacity that may be there.

 

[Geo]

Jen-Hsun Huang on the rumor:

It's hard to say whether the rumor is true or not, but there are some things here that are logical and there are some things that are not. It's hard to invest multiple hundreds of millions of dollars per generation when you don’t generate multiple hundreds of millions of dollars of profit for that particular product line. Over time, I think it becomes increasingly difficult to sustain innovation at that level.

On the other hand, AMD's strategy as an alternative to Intel (nasdaq: INTC - news - people ) is to be the open ecosystem [and] to partner with all of the leaders in the marketplace, whereas Intel has a closed strategy with Centrino and Viiv [which use only Intel products]. If AMD decides to [buy ATI], it closes the system pretty dramatically. Following a giant with a giant's strategy is not always the most clever thing.

http://www.forbes.com/technology/2006/06/22/ps3-phone-nvidia_cx_ck_0622nvidia.html?partner=rss

Props to epicstruggle for the tip on the interview. . .